A background-free phenology index for improved monitoring of vegetation phenology

Abstract

Accurate monitoring of vegetation phenology (e.g., the start and end of the growing season, SOS and EOS) is helpful for understanding the impacts of climate change on vegetation and the terrestrial carbon cycle. The remote sensing-based vegetation index (e.g., enhanced vegetation index, EVI) and remote sensing-based phenology index (e.g., normalized difference greenness index, NDGI) are the major data sources for phenology monitoring at regional and global scales. However, these remote sensing-based indices are vulnerable to the influences of backgrounds and their variations. As a result, it is difficult to obtain high-precision vegetation phenology by using only the remote sensing-based indices, especially for the EOS. In this study, we developed a background-free phenology index (BFPI) by coupling the remote sensing-based index and the meteorological factor-based normalized growing season index (calculated by the normalized daily minimum temperature, vapor pressure deficit, and photoperiod). The BFPIs (BFPIEVI and BFPINDGI) were constructed as products of the EVI/NDGI and normalized growing season index. The performances of the BFPIs in phenology monitoring were evaluated by using the gross primary production data from 64 carbon flux towers and green chromatic coordinate data from 57 PhenoCam sites in forests and grasslands in the Northern Hemisphere. The results showed that the BFPIs performed better than the remote sensing-based indices in phenology monitoring for both forests and grasslands. The BFPINDGI performed better than the BFPIEVI for SOS monitoring, and the two BFPIs had nearly equal performances for monitoring the EOS of grasslands. For forests, the BFPIEVI performed better than the BFPINDGI in phenology monitoring. Although the performances of the BFPIs were limited for EOS monitoring, the phenology monitoring accuracy based on BFPIs were still obviously higher than that based on the remote sensing indices. Overall, the newly-developed BFPI that integrates biological and meteorological factors not only improved the precision of phenology monitoring, but also provided a new perspective for multisource data-based phenology monitoring.